Abstract Little empirical evidence exists as to the behavior of customers as they self-select from usage-senstive to flat rate tariffs. Since this study was performed prior to service availability, neither choice behavior nor usage stimulation were observable. The customer's survey response to a price query (‘would you purchase EAS at price $x’) is observed, however. We assume the customer is at least implicitly comparing his ‘projected’ benefits as measured by consumer surplus to the subscription price. By assuming an underlying demand for toll use, we can calculate the consumer surplus and derive the new usage level in terms of the initial or observed usage quantity. This framework provides a basis for a probabilistic choice model and allows joint prediction of penetration levels and usage stimulation rates. Customer demographics are introduced through the price parameter from the toll demand equation which relates stimulated usage to the initial usage level as well as in the choice parameters directly. Assuming a logistic framework for choice leads to rather a straight-forward maximum likelihood estimation problem. The resulting coefficient estimates are used to predict development and stimulation rates at various price levels.